Deep Generative Methods for Producing Forecast Trajectories in Power Systems

With the expansion of renewables in the electricity mix, power grid variability will increase; hence, the system needs to be robust to guarantee its security. To anticipate the design of new tools and processes, Transmission System Operators (TSOs) must conduct analyses to simulate the future operation of power systems. Then, these simulations are used as inputs in decision-making processes. In this context, we investigate using deep learning models to generate energy production and load forecast trajectories. We adapt autoregressive networks and normalizing flows to capture the spatiotemporal correlations in these multivariate time series, demonstrating their effectiveness against the current copula-based statistical approach. We conduct extensive experiments on the French TSO RTE wind forecast data and compare the different models with ad hoc evaluation metrics for time series generation.